Fix memleak in TorController [rework]

It looks like, TorController::disconnected_cb(TorControlConnection&
conn) gets called multiple times which results in multiple event_new().

Avoid this by creating the event only once in the constructore, and
deleting it only once in the destructor (thanks to Cory Fields for the
idea).

Replaces the fix by Jonas Schnelli in #7610, see discussion there.

1 files changed, 9 insertions, 4 deletions

diff --git a/src/torcontrol.cpp b/src/torcontrol.cpp
index 10170dbce..a04b5c302 100644
--- a/src/torcontrol.cpp
+++ b/src/torcontrol.cpp
@@ -398,6 +398,9 @@ TorController::TorController(struct event_base* base, const std::string& target)
     target(target), conn(base), reconnect(true), reconnect_ev(0),
     reconnect_timeout(RECONNECT_TIMEOUT_START)
 {
+    reconnect_ev = event_new(base, -1, 0, reconnect_cb, this);
+    if (!reconnect_ev)
+        LogPrintf("tor: Failed to create event for reconnection: out of memory?\n");
     // Start connection attempts immediately
     if (!conn.Connect(target, boost::bind(&TorController::connected_cb, this, _1),
          boost::bind(&TorController::disconnected_cb, this, _1) )) {
@@ -413,8 +416,10 @@ TorController::TorController(struct event_base* base, const std::string& target)
 
 TorController::~TorController()
 {
-    if (reconnect_ev)
-        event_del(reconnect_ev);
+    if (reconnect_ev) {
+        event_free(reconnect_ev);
+        reconnect_ev = 0;
+    }
     if (service.IsValid()) {
         RemoveLocal(service);
     }
@@ -626,8 +631,8 @@ void TorController::disconnected_cb(TorControlConnection& conn)
 
     // Single-shot timer for reconnect. Use exponential backoff.
     struct timeval time = MillisToTimeval(int64_t(reconnect_timeout * 1000.0));
-    reconnect_ev = event_new(base, -1, 0, reconnect_cb, this);
-    event_add(reconnect_ev, &time);
+    if (reconnect_ev)
+        event_add(reconnect_ev, &time);
     reconnect_timeout *= RECONNECT_TIMEOUT_EXP;
 }